US accuses six Chinese AI firms of ‘industrial-scale’ trade-secret theft
US officials allege that six Chinese companies used model distillation to train their AI systems, amid wider tensions over access to advanced technology.
US officials have accused six Chinese artificial intelligence firms of “industrial-scale” theft of US AI trade secrets, according to reporting by MIT Technology Review.
The named companies are DeepSeek, Moonshot AI, Alibaba and Z.AI. The supplied material does not identify the other two firms.
Officials reportedly allege that the companies used model distillation to help train their systems. The technique involves using outputs from an existing AI model to assist in training another system; the material does not establish that its use is unlawful.
The alleged targets reportedly include models developed by Anthropic, OpenAI, Google and xAI. However, the precise companies involved and the basis for attributing the alleged activity to each firm remain unclear.
The accusations come amid wider US–China tensions over artificial intelligence capability and access to advanced systems. The supplied material does not provide evidence substantiating the allegations or detail any official enforcement action.
